Determinants of Mortality of Patients with COVID-19 in Wuhan, China: a Case-Control Study

3950

Original Article

Determinants of mortality of patients with COVID-19 in Wuhan, China: a case-control study

Jian Li1#^, Luyu Yang2#, Qian Zeng3,4#, Qingyun Li5, Zhitao Yang6, Lizhong Han3,4, Xiaodong Huang7, Erzhen Chen6

1Clinical Research Center, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China; 2Department of ICU/Emergency, Wuhan Third Hospital, Tongren Hospital of Wuhan University, Wuhan, China; 3Department of Laboratory Medicine, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China; 4Department of Clinical Microbiology, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China; 5Department of Respiration and Critical Care Diseases, Shanghai Jiao Tong University School of Medicine, Shanghai, China; 6Emergency Department, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China; 7Department of Gastroenterology, Wuhan Third Hospital, Tongren Hospital of Wuhan University, Wuhan, China Contributions: (I) Conception and design: E Chen, L Han, J Li; (II) Administrative support: E Chen, X Huang; (III) Provision of study materials or patients: L Yang, Q Li, Z Yang; (IV) Collection and assembly of data: All authors; (V) Data analysis and interpretation: J Li, Q Zeng; (VI) Manuscript writing: All authors; (VII) Final approval of manuscript: All authors. #These authors contributed equally to this work. Correspondence to: Erzhen Chen. Emergency Department, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China. Email: [email protected]; Lizhong Han. Department of Laboratory Medicine, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China; Department of Clinical Microbiology, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China. Email: [email protected]; Xiaodong Huang. Department of Gastroenterology, Wuhan Third Hospital, Tongren Hospital of Wuhan University, Wuhan, China. Email: [email protected].

Background: Coronavirus disease 2019 (COVID-19) has resulted in an overwhelmed challenge to the healthcare system worldwide. Methods: A case-control study of COVID-19 patients in Wuhan Third Hospital was conducted. 96 deceased COVID-19 patients and 230 discharged patients were included as the case group and control group, respectively. Demographic, epidemiological, clinical and laboratory variables on admission were collected from electronic medical records. Univariate and multivariate logistic regression were adopted to investigate the independent predictors of mortality. A nomogram was formed for predicting the mortality risk. Results: The multivariate stepwise logistic model demonstrated that age of 60+ years (OR =4.426, 95% CI: 1.955–10.019), comorbidity of cerebrovascular disease (OR =7.084, 95% CI: 1.545–32.471), white blood cell (WBC) count >9.5×109/L (OR =7.308, 95% CI: 1.650–32.358), platelet count <125×109/L (OR =5.128, 95% CI: 2.157–12.191), aspartate aminotransferase (AST) >40 U/L (OR =2.554, 95% CI: 1.253–5.206), cystatin C >1.1 mg/L (OR =4.132, 95% CI: 2.118–8.059), C reactive protein (CRP) ≥100 mg/L (OR =2.830, 95% CI: 1.311–6.109), creatine kinase isoenzymes (CK-MB) >24 U/L (OR =6.015, 95% CI: 2.119–17.07) and D-dimer >5 μg/L (OR =4.917, 95% CI: 1.619–14.933) were independent predictors of mortality of COVID-19 patients. The nomogram demonstrated a well discriminatory accuracy for mortality prediction with a C-index of 0.903. Conclusions: The determinants identified may help to determine patients at high risk of death at an early stage and guide the optimal treatment.

Keywords: Coronavirus disease 2019 (COVID-19); mortality; determinant; case control study

Submitted Oct 25, 2020. Accepted for publication Jan 21, 2021. doi: 10.21037/apm-20-2107 View this article at: http://dx.doi.org/10.21037/apm-20-2107

^ ORCID: 0000-0003-0901-6458.

Introduction Guideline for the Diagnosis and Treatment of COVID-19 issued by the National Health Commission of the People’s Since December 2019, the pandemic of coronavirus disease Republic of China (v.5) as reported previously (14), and were 2019 (COVID-19), which originated from infection of laboratory-confirmed COVID-19. All deceased patients were severe acute respiratory syndrome coronavirus 2 (SARS- recruited into this study as the case group. Two discharged CoV-2), has spread around the world rapidly and raised patients to each deceased individual were randomly sampled serious concerns (1,2). COVID-19 has given rise to to be used as controls. Seven deceased patients were excluded an overwhelming destructive impact worldwide. As of from the study owing to non-accessibility of important 30 August 2020, the World Health Organization (WHO) laboratory findings and data of comorbidities, and 230 has reported a cumulative total of nearly 25 million discharged individuals were included in the study eventually. COVID-19 cases and 800,000 deaths globally (3). Regardless The Ethics Committees of Wuhan Third Hospitals has of the fact that the majority of confirmed patients were approved this study (No.: KY2020-007) and waived the considered to be mild or moderate, 5% were critical and individual informed consent due to the retrospective analysis. 14% severe on the basis of the largest survey of 72,314 The study was carried out in conformity to the Declaration patients to date (4). As with SARS-CoV (5) and Middle of Helsinki (as revised in 2013). East Respiratory Syndrome coronavirus (MERS-CoV) (6), COVID-19 is more easily predisposed to acute respiratory distress syndrome, respiratory failure, heart failure, Date collection septic shock and even death among susceptible cases (7). Data were gathered from April 28 to May 7, 2020. The Accumulated evidence from the latest studies indicates research team reviewed the medical records of COVID-19 the risk factors concerning fatality of COVID-19 are cases. A standardized case report form was adopted to collect heterogeneous and may include male, older age, comorbidity, the data of demographics, epidemiology, symptoms and signs, smoking, elevated D-dimer level, coagulopathy, neutrophilia laboratory and complication from electronic medical records, and organ dysfunction (8-13). However, most of those early and two experienced doctors cross-checked all of data. In case studies came from cases series and were descriptive analysis of unavailable epidemiological data or medical history from without consideration of adequate adjustment for potential medical records, researchers interviewed the patients or their confounding factors. So the estimations of risk factors of close relatives to further ascertain. If there was a discrepancy mortality were not very robust. Studies that address these between the medical records and the subjective description, limitations are necessary in order to explore the determinants data from medical records were used. Chest CT scan were of death for COVID-19 individuals. Therefore, a case- excluded from the analysis due to massive missing of CT data control study based on COVID-19 patients from 2 campuses at an early stage of the epidemic. of a designated hospital in Wuhan, Hubei province was conducted to identify the determinants of mortality of patients with COVID-19. We present the following article in Laboratory procedures accordance with the STROBE reporting checklist (available Laboratory method for confirming SARS-CoV-2 was at http://dx.doi.org/10.21037/apm-20-2107). reported elsewhere (1). Specifically, the Chinese Center for Disease Control and Prevention (CDC) and local CDC Methods were responsible for pathogen detecting. The real-time reverse transcription polymerase chain reaction assay (RT- Study design and participants PCR) was adopted to detect SARS-CoV-2 in oropharyngeal This case-control study enrolled COVID-19 patients with swab samples. The discharge criteria involved remission definite outcomes from 2 campuses (Shouyi campus and of respiratory symptoms, absence of fever for 3 days and Guanggu campus) of Wuhan Third Hospital. A total of above, completely improvement in bilateral lungs by chest 1,574 COVID-19 patients (1,241 in Guanggu and 334 in computed tomography (CT), and negative for 2 times in Shouyi) were admitted to this designated hospital from oropharyngeal swab specimens for SARS-CoV-2 RNA at 8 January 2020 to 17 March 2020. One hundred and three of least 1 day apart. Initial clinical laboratory examinations them (6.54%) were deceased and the rest of 1471 individuals included serum biochemical tests (including liver and were discharged. All patients were diagnosed following the kidney functions), complete blood count, coagulation

© Annals of Palliative Medicine. All rights reserved. Ann Palliat Med 2021;10(4):3937-3950 | http://dx.doi.org/10.21037/apm-20-2107 Annals of Palliative Medicine, Vol 10, No 4 April 2021 3939 profile, myocardial enzymes and D-dimer. Frequency of 96 cases had died during hospitalization, and 230 cases had examinations was placed under the discretion of doctors. recovered and been discharged. The median age of the 326 cases was 62 years (IQR 50–71; range, 25 to 98 years), and male accounted for 52.5% (Table 1); 133 (40.8%) cases Statistical analysis presented at least one comorbidity, with hypertension Continuous variables were expressed as median with being the most common comorbidity (40.9%), followed by interquartile range (IQR) and the Mann-Whitney U test was diabetes (20.1%) and cardiovascular disease (9.0%). Only adopted to compare the difference between deceased group 9 (3.2%) cases had a current smoking. Fever (76.8%) and and discharged group. Categorical variables were presented dry cough (71.1%) were the most common symptoms on as the frequency with percentages, and were analyzed by the admission, followed by chest tightness (29.6%) and sputum Pearson Chi-Square test or Fisher's Exact Test as appropriate. production (14.8%). Median systolic blood pressure and Potential risk factors associated with death comprised of the median diastolic blood pressure on admission were 130 and following features at admission: demography and epidemiology, 80 mmHg, respectively. comorbidities, symptoms and signs, together with laboratory Table 1 showed the substantial differences of laboratory results. To explore the determinants of COVID-19 mortality, findings on admission between deceased group and the univariate logistic regression was used to compare the discharged group. The results demonstrated that remarkably characteristics between case group and control group. A higher proportion of cases showing unnormal aspartate backward stepwise multivariate logistic regression analysis aminotransferase (AST), and significantly higher levels (both of entry and removal probability were 0.05) was further of C reactive protein (CRP), white blood cell (WBC) performed excluding variables which were not significant in count, neutrophil count, neutrophil/lymphocyte ratio univariate logistic analysis. Multiple imputation for missing (NLR), globulin, total bilirubin, blood urea nitrogen, data of some cases at admission was performed if missing rate creatinine, cystatin C, prothrombin time, creatine kinase, were less than 20% (the maximum missing rate of variable α-hydroxybutyrate dehydrogenase, CK-MB and D-dimer was 16.87%). We utilized the fully conditional specification were observed in the deceased group than in the discharged (FCS) regression methods to impute continuous variables and group, whereas, the platelet count, levels of lymphocyte the discriminant function method to impute the classification count and thrombin time were remarkably lower in the variables. Continuous variables of laboratory findings were decreased group as compared with the discharged group converted into categorical variables with their reference values (P<0.05). There was not any statistically significant difference as threshold before logistic regression model was fitted. All with regard to monocyte count, alanine aminotransferase tests were conducted two-sided at the 5% significance level. All (ALT) and total protein between the decreased group and above statistical analyses were performed using SAS software the discharged group (P>0.05). Table S1 presented the (v. 9.4) (SAS Institute Inc., USA). We constructed a nomogram normal ranges of laboratory indicators. Totally, the most on the basis of determinants identified in the multivariate commonly observed complication was respiratory failure logistic model using rms package in R Project v.3.5.2 (The R (75.5%), followed by heart failure (30.9%), acute respiratory Foundation for Statistical Computing, Vienna, Austria. http:// distress syndrome (11.3%), acute cardiac injury (10.4%) and acute kidney injury (9.3%). No significant differences www.r-project.org). The concordance index (C-index) was between deceased group and discharged group were observed used to measure the predictive performance of the nomogram. with regard to all kinds of complication (Table 2). Overall, The C-index fluctuated in the interval of 0.5 and 1.0 with the median length of hospitalization (LOH) was 15 days 1.0 representing a totally correct discrimination and 0.5 (IQR 9–23) and the median LOH of deceased group was denoting random chance. remarkably shorter than that of discharged group (P<0.05).

Results Univariate analysis of risk factors for COVID-19 mortality Demographics, clinical, laboratory findings and The univariate logistic analysis of possible risk factors is complications shown in Table 3. Demographic characters including age Among 326 COVID-19 patients (274 in Guanggu campus ≥60 years old (unadjusted OR =3.846) and male (unadjusted and 52 in Shouyi campus) who had definite outcome, OR =1.977) were associated with increased risk of death.

Table 1 Demographic, clinical characters and laboratory parameters on admission of 326 COVID-19 patients in Wuhan, China Group Variable All (n=326) P value Dead (n=96) Discharged (n=230)

Demographics, comorbidities and clinical characteristics

Age, yrs 62 [50–71] 70.5 [62–80] 57 [44–67] <0.0001

Male 171/326 (52.5) 63/96 (65.6) 108/230 (47.0) 0.003

Exposure to confirmed COVID-19 27/284 (9.5) 5/78 (6.4) 22/206 (10.7) 0.274 patients

Current smoking 9/280 (3.2) 0/76 (0.0) 9/204 (3.2) 0.119

Hypertension 121/296 (40.9) 49/84 (58.3) 72/212 (34.0) <0.0001

Diabetes 59/293 (20.1) 24/81 (29.6) 35/212 (16.5) 0.015

Cardiovascular disease 26/288 (9.0) 8/78 (10.3) 18/210 (8.6) 0.818

COPD 4/283 (1.4) 2/76 (2.6) 2/207 (1.0) 0.638

Carcinoma 7/284 (2.5) 6/78 (7.7) 1/206 (0.5) 0.002

Cerebrovascular disease 10/287 (3.5) 8/80 (10.0) 2/207 (1.0) 0.001

Fever 218/284 (76.8) 59/78 (75.6) 159/206 (77.2) 0.875

Dry cough 202/284 (71.1) 50/78 (64.1) 152/206 (73.8) 0.142

Sputum production 42/284 (14.8) 19/78 (24.4) 23/206 (11.2) 0.008

Myalgia 21/284 (7.4) 5/78 (7.4) 16/206 (7.8) 0.804

Chest tightness 84/284 (29.6) 25/78 (32.1) 59/206 (28.6) 0.662

Dyspnea 10/275 (3.6) 5/73 (6.8) 5/202 (2.5) 0.137

Systolic blood pressure 130 [120–141] 134 [122–150] 130 [120–140] 0.091

Diastolic blood pressure 80 [74–87] 80 [75–89] 80 [74–86] 0.580

Heart rate 83 [80–96] 85.5 [80–102] 82 [80–93.8] 0.073

Transcutaneous oxygen saturation 96.5 [89.0–99.4] 91. [65.7–99.1] 97.4 [94.9–99.7] 0.067

Laboratory findings

CRP, mg/L 34.6 [9.8–83.4] 83.1 [36.1–162.5] 21.0 [5.1–58.1] <0.0001

WBC count, ×109/L 5.1 [4.0–7.0] 6.4 [4.4–10.6] 4.8 [3.7–6.3] <0.0001

Neutrophil count, ×109/L 3.57 [2.43–5.34] 5.25 [3.135–9.055] 3.26 [2.245–4.48] <0.0001

Lymphocyte count, ×109/L 0.91 [0.68–1.33] 0.79 [0.43–1.06] 1.01 [0.75–1.46] <0.0001

Monocyte count, ×109/L 0.36 [0.23–0.49] 0.38 [0.24–0.49] 0.36 [0.23–0.49] 0.774

Platelet count, ×109/L 182.0 [135–241.7] 159.6 [120–205] 191 [142.23–253.7] <0.0001

Neutrophil/lymphocyte ratio 3.5 [2.2–6.5] 6.6 [3.8–15.2] 3.0 [2.0–4.6] <0.0001

ALT, U/L 27 [16.8–43.0] 32 [16–54] 26 [17–40] 0.186

AST, U/L 32 [22.5–49.5] 45 [26–79] 30 [21–40] <0.0001

Total protein, g/L 66.2 [62.7–70.1] 65.8 [60.3–69.7] 66.55 [62.8–70.4] 0.116

Globulin, g/L 30.4 [26.8–33.5] 32 [28.1–35.7] 29.7 [26.3–32.6] <0.0001

Table 1 (continued)

Table 1 (continued) Group Variable All (n=326) P value Dead (n=96) Discharged (n=230)

Total bilirubin, mmol/L 9.3 [6.7–12.7] 10.3 [7.4–14.7] 8.9 [6.6–12.3] 0.013

Blood urea nitrogen, mmol/L 4.6 [3.4–7.2] 8.1 [5.5–16.1] 3.9 [3.2–5.2] <0.0001

Creatinine, umol/L 69.8 [54.7–91.6] 82.1 [63.9–128] 65.3 [53.8–82.8] <0.0001

Cystatin C, mg/L 1.0 [0.8–1.3] 1.3 [1–2.0] 0.9 [0.7–1.1] <0.0001

PT 11.8 [11.4–12.5] 12.4 [11.6–13.4] 11.6 [11.3–12.2] <0.0001

TT 20.3 [18.7–22.6] 19.7 [18.5–21.9] 20.5 [18.8–23.4] 0.041

Creatine kinase, U/L 88 [53.5–173] 132 [64–364] 77 [55–137] <0.0001

α-HBDH, IU/L 206 [160–291.3] 286 [206–354] 187 [155–255] <0.0001 CK-MB, U/L 12 [8–17] 18 [12–26] 11 [8–13] <0.0001

D-dimer, μg/L 0.81 [0.41–2.37] 2.53 [0.83–5.45] 0.60 [0.35–1.13] <0.0001 Data are presented as median [IQR] or n/total (%). P value means the comparison between deceased group and discharged group. COPD, chronic obstructive pulmonary disease; CRP, C-reactive protein; WBC, white blood cell; ALT, alanine aminotransferase; AST, aspartate aminotransferase; PT, prothrombin time; TT, thrombin time; α-HBDH, α-hydroxybutyrate dehydrogenase; CK-MB, creatine kinase isoenzyme.

Table 2 Complications and prognosis of 326 COVID-19 patients in Wuhan, China Group, n/total (%) Variable All (n=326), n/total (%) P value Dead (n=96) Discharged (n=230)

Type I respiratory failure 74/98 (75.5) 65/86 (75.6) 9/12 (75.0) 1.000

ARDS 11/97 (11.3) 11/85 (12.9) 0/12 (0.0) 0.349

Sepsis 5/97 (5.2) 5/85 (5.9) 0/12 (0.0) 1.000

Septic shock 7/97 (7.2) 6/85 (7.1) 1/12 (8.3) 1.000

Heart failure 30/97 (30.9) 29/85 (34.1) 1/12 (8.3) 0.097

Acute cardiac injury 10/96 (10.4) 10/84 (11.9) 0/12 (0.0) 0.353

Acute kidney injury 9/97 (9.3) 9/85 (10.6) 0/12 (0.0) 0.596

Acute liver injury 2/97 (2.1) 2/85 (2.4) 0/12 (0.0) 1.000

Acidosis 6/97 (6.2) 6/85 (7.1) 0/12 (0.0) 1.000

Alkalosis 1/96 (1.0) 1/84 (1.2) 0/12 (0.0) 1.000

Length of hospitalization, days 15 [9–22] 9 [5–13.8] 17 [12–25] <0.0001 P value means the comparison between deceased group and discharged group. ARDS, acute respiratory distress syndrome.

Comorbidities involving hypertension (OR =2.962), diabetes from illness onset to admission (OR =1.028) were found to be (OR =2.120), carcinoma (OR =18.007) and cerebrovascular related to mortality. Most of laboratory findings on admission disease (OR =4.655) appeared to be susceptible to mortality. including elevated CRP (OR =6.363), elevated WBC count Clinical symptoms including sputum production (OR =2.647) (OR =15.862), elevated NLR (OR =1.137), elevated ALT (OR and dyspnea (OR =4.192), together with prolonged interval =2.256), elevated AST (OR =3.259), elevated total bilirubin

Table 3 Univariate analysis of risk factors for mortality of COVID-19 patients in Wuhan, China Variable Level β Unadjusted OR 95% CI P value Demographics, comorbidities and clinical characteristics

Age (years) <60 Ref

≥60 1.347 3.846 2.187–6.766 0.0001

Gender Male 0.681 1.977 1.206–3.240 0.0069

Female Ref

Exposure to confirmed patient Present –0.151 0.860 0.369–2.004 0.7260

Absent Ref

Current smoking Present –0.859 0.424 0.092–1.948 0.2699

Absent Ref

Hypertension Present 1.086 2.962 1.812–4.840 <0.0001

Absent Ref

Cardiovascular disease Present 0.521 1.683 0.777–3.645 0.1869

Absent Ref

Diabetes Present 0.752 2.120 1.217–3.695 0.008

Absent Ref

Carcinoma Present 2.891 18.007 2.185–148.430 0.0072

Absent Ref

COPD Present 1.601 4.956 0.892–27.529 0.0673

Absent Ref

Cerebrovascular disease Present 1.538 4.655 1.518–14.280 0.0072

Absent Ref

Fever Absent –0.107 0.898 0.516–1.564 0.7047

Present Ref

Dry cough Absent –0.217 0.805 0.483–1.342 0.4056

Present Ref

Sputum production Absent 0.974 2.647 1.442–4.860 0.0017

Present Ref

Myalgia Absent 0.325 1.384 0.589–3.250 0.4560

Present Ref

Chest tightness Absent 0.304 1.355 0.810–2.267 0.2475

Present Ref

Dyspnea Absent 1.433 4.192 1.748–10.056 0.0013

Present Ref

Interval from illness onset to admission 0.028 1.028 1.002–1.055 0.032

Table 3 (continued)

Variable Level β Unadjusted OR 95% CI P value Laboratory findings

CRP, mg/L ≥100 1.850 6.362 3.606–11.224 <0.0001

<100 Ref

WBC count, ×109/L <3.5 Ref

3.5–9.5 0.460 1.585 0.753–3.333 0.225

>9.5 2.764 15.862 5.588–45.021 <0.0001

Neutrophil count, ×109/L ≥1.8 Ref

<1.8 –0.648 0.523 0.208–1.315 0.1682

Lymphocyte count, ×109/L <1.1 0.993 2.699 1.558–4.677 0.0004

≥1.1 Ref

Monocyte count, ×109/L >0.8 0.615 1.850 0.624–5.484 0.2669

≤0.8 Ref

Platelet count, ×109/L <125 1.022 2.778 1.560–4.948 0.0005

≥125 Ref

NLR 0.129 1.137 1.084–1.193 <0.0001

ALT, U/L >50 0.814 2.256 1.270–4.009 0.0055

≤50 Ref

AST, U/L >40 1.182 3.259 1.971–5.390 <0.0001

≤40 Ref

Total bilirubin, mmol/L >20.5 1.046 2.847 1.167–6.948 0.0215

≤20.5 Ref

Total protein, g/L <60 1.438 4.214 2.111–8.411 <0.0001

≥60 Ref

Globulin, g/L >30 0.723 2.060 1.260–3.369 0.0040

≤30 Ref

Blood urea nitrogen, mmol/L >9.5 2.313 10.108 5.132–19.909 <0.0001

≤9.5 Ref

Creatinine, umol/L >111 1.256 3.512 1.872–6.587 <0.0001

≤111 Ref

Cystatin C, mg/L >1.1 1.870 6.489 3.847–10.947 <0.0001

≤1.1 Ref

PT, second >16 –0.399 0.671 0.215–2.093 0.4918

≤16 Ref

TT, second >25 –0.680 0.506 0.226–1.137 0.0990

Table 3 (continued)

Variable Level β Unadjusted OR 95% CI P value ≤25 Ref

α-HBDH, IU/L >182 1.011 2.747 1.586–4.760 0.0003 ≤182 Ref

Creatine Kinase, U/L >200 1.051 2.860 1.656–4.941 0.0002

≤200 Ref

CK-MB, U/L >24 1.745 5.724 2.713–12.076 <0.0001

≤24 Ref

D-dimer, μg/L >5 2.158 8.654 3.686–20.315 <0.0001

0.5–5 1.572 4.816 2.336–9.929 <0.0001

<0.5 Ref COPD, chronic obstructive pulmonary disease; CRP, C-reactive protein; WBC, white blood cell; NLR, Neutrophil/lymphocyte ratio; ALT, alanine aminotransferase; AST, aspartate aminotransferase; PT, prothrombin time; TT, thrombin time; α-HBDH, α-hydroxybutyrate dehydrogenase; CK-MB, creatine kinase isoenzyme.

(OR =2.847), elevated globulin (OR =2.060), elevated blood those of less than 60 years. Patients with cerebrovascular urea nitrogen (OR =10.108), elevated creatinine (OR =3.512), disease had 7.084 (95% CI: 1.545–32.471) times greater risk elevated cystatin C (OR =6.489), elevated α-hydroxybutyrate of death compared with patients without cerebrovascular dehydrogenase (OR =2.747), elevated creatine kinase (OR disease. The risk of mortality was found higher for patients =2.860), elevated CK-MB (OR =5.724), elevated D-dimer with WBC count >9.5×109/L (OR =7.308; 95% CI: 1.650– (OR =4.816 for 0.5–5 μg/L, OR =8.654 for >5 μg/L), 32.358) when compared with those having WBC count and decreased lymphocyte count (OR =2.699), decreased <3.5×109/L. The risk of mortality was found significantly platelet count (OR =2.778), together with decreased total higher for patients with lower platelet count (<125×109/L) protein (OR =4.214) looked like to be apt to mortality. In than those with platelet count ≥125×109/L with OR contrast, exposure to confirmed patient, current smoking, being 5.128 (95% CI: 2.157–12.191). COVID-19 cases comorbidities of cardiovascular disease, COPD, fever, cough, with AST >40 U/L on admission had a 2.554 (95% CI: myalgia, chest tightness, neutrophil count, monocyte count, 1.253–5.206)-fold higher risk of mortality when comparison PT and TT were not found to be statistically different was performed with patients having normal AST level. between 2 groups (P>0.05). Patients with cystatin C >1.1 mg/L at admission had a 4.132 (95% CI: 2.118–8.059)-fold greater risk of mortality when comparison was performed with those having cystatin C Multivariate analysis of determinants of COVID-19 count <1.1 mg/L. Table 3 also showed that patients with mortality CRP ≥100 mg/L and patients with CK-MB >24 U/L Incorporating 26 significant predictors for death in had 2.830 (95% CI: 1.311–6.109) and 6.015 (95% CI: univariable analysis into a multivariable logistic model with 2.119–17.072) times greater risk of mortality compared multiordinal variables converted into dummy variables, with those opposite, respectively. Notably, patients with and the final model retained the 9 variables which were elevated D-dimer were more likely to die with OR for level independently significant determinants of COVID-19 of between 0.5–5 μg/L and more than 5 μg/L being 2.233 mortality (Table 4). Collinearity diagnostics of independent (95% CI: 0.922–5.406) and 4.917 (95% CI: 1.619–14.933), variables did not show multicollinearity. The results of respectively. multivariate analysis demonstrated that the mortality risk All the independent predictors for mortality of was 4.426 (95% CI: 1.955–10.019) times greater among COVID-19 individuals in the data set were incorporated cases belonging to the age group 60+ years as compared with into the nomogram fitting.Figure 1 illustrates the predictive

Table 4 Multiple logistic regression analysis of determinants for COVID-19 mortality in Wuhan Variable Level β OR 95% CI P value Age (years) ≥60 1.488 4.426 1.955–10.019 0.0004

<60 Ref

Cerebrovascular disease Present 1.958 7.084 1.545–32.471 0.0177

Absent Ref

WBC count (×109/L) <3.5 Ref

3.5–9.5 –0.067 0.936 0.328–2.665 0.9007

>9.5 1.989 7.308 1.650–32.358 0.0088

Platelet count, ×109/L <125 1.635 5.128 2.157–12.191 0.0002

≥125 Ref

AST, U/L >40 0.938 2.554 1.253–5.206 0.0099

≤40 Ref

Cystatin C, mg/L >1.1 1.419 4.132 2.118–8.059 <0.0001

≤1.1 Ref

CRP –mg/L ≥100 1.040 2.830 1.311–6.109 0.0081

<100 Ref

CK-MB, U/L >24 1.794 6.015 2.119–17.072 0.0007

≤24 Ref

D-dimer, μg/L >5 1.593 4.917 1.619–14.933 0.0050

0.5–5 0.803 2.233 0.922–5.406 0.0750

<0.5 Ref WBC, white blood cell; AST, aspartate aminotransferase; CRP, C-reactive protein; CK-MB, creatine kinase isoenzyme.

nomogram established for the mortality risk probability. A currently. A recent systematic review of predicting models COVID-19 patient’s mortality risk can simply be calculated conceived that the performance of prognostic evaluation of by summing the scores for individual factor on the points COVID-19 may be misleading and overoptimistic, because scale. The nomogram demonstrated a well discriminatory of the great risk of bias in vague outcome definition and accuracy for mortality prediction with a C-index of 0.903. subject selection (15). Regardless of the fact that there were several retrospective cohort studies evaluating mortality risk score in COVID-19 patients which aimed to earlier Discussion identify mortality risk (13,16,17), they were subject to the Although the COVID-19 pandemic is probable to have selection bias of patients. The current case-control study initiated from a zoonotic transmission event, it is well- was designed to determine the risk factors of mortality of known that SARS-CoV-2 is highly pathogenic in humans. COVID-19 patients in Wuhan, the source of the epidemic, Mortality is the most vital problem when dealing with in 2020. Although similar in terms of objectives to the epidemics in the face of the gradually scarce health above retrospective cohort studies, our study randomly resources. Early identifying determinant of death of selected the controls from the discharged patients, which COVID-19 contributes to allocation of limited critical care was different from previous cohort studies and apparently and becomes the priority to minimize fatality. Nevertheless, reduced the bias in patient selection. Compare to prior there is a lack of reliable models for mortality predicting similar reports, our study added some new information

0 10 20 30 40 50 60 70 80 90 100 Points

≥60 age

<60 Yes CVD No <3.5 WBC 3.5−9.5 >9.5 <125 Platelet ≥125 >40 AST ≤40 >1.1 Cystatin.C ≤1.1 >100 CRP <100 >24 CKMB ≤24 0.5−5 Ddimer <0.5 >5 Total Points 0 50 150 250 350 450 550 Mortality Risk 0.1 0.3 0.5 0.7 0.9

Figure 1 Nomogram for predicting mortality of COVID-19 patient. The nomogram is employed by summing the points identified on the scale for 9 factors to acquire the total points, and draw a vertical line downward to the mortality risk axis to determine the probability of death. CVD, cerebrovascular disease; WBC, white blood cell; AST, aspartate aminotransferase; CRP, C-reactive protein; CKMB, creatine kinase isoenzyme. on independent risk factors associated with higher odds of that commorbidity of cerebrovascular disease predisposed mortality of COVID-19 inpatients such as comorbidity of to death in COVID-19 patients. The SARS-CoV infection cerebravascular disease, higher cystatin C and WBC count has been reported in the brains from both experimental on admission, except for already described risk factors animals and cases, where the heavily infected brainstem was including older age, lower platelet count, higher AST, CRP, observed (20). Considering the similarities between SARS- CK-MB and D-dimer on admission in the literatures. CoV-2 and SARS-CoV in various aspects, cases with Evidence is increasingly accumulated with respect to cerebrovascular comorbidity are more liable to be infected the determinants of COVID-19 mortality. In accordance by SARS-CoV-2 due to low immunity and to develop with recent studies, advanced age has been recorded as a cerebral complications, which is strongly associated with determinant of mortality in COVID-19 cases (4 8,13,16,17), lethal outcome. Although leucopenia is more common in which is proved by our study. A deficiency in controlling COVID-19 patients at an early stage, our study revealed for viral replication and more prolonged proinflammatory that greater WBC count at admission was a dependent response may be caused by the excess expression of type determinant of COVID-19 mortality. We speculated 2 cytokines and the age-dependent defects in T-cell and these patients with an elevated WBC count might get the B-cell function, potentially leading to a poor prognosis second infection with bacteria, which accelerated disease or even death (18). Previous studies showed COVID-19 progression and was more likely to induce death. It has patients combined with basic disease such as hypertension been reported that COVID-19 had a significant impact and diabetes tended to develop the adverse outcomes (7,19). on the liver function (21,22). A study incorporating 3,428 Less consistent with these studies, our finding suggested COVID-19 cases showed that higher serum levels of ALT

(mean difference =7.35 U/L, P<0.001) and AST (mean (7,19), which is proven by our research. We found patients difference =8.84 U/L, P<0.001) were related to a remarkable with elevated D-dimer were associated with fatal outcome. deterioration in the severity of COVID-19 and death (23). Particularly, patients with D-dimer greater than 5 μg/L Our univariate analysis also revealed cases with elevated ALT on admission showed a 4.917 (95% CI: 1.619–14.933)- and AST had 2.256- and 3.259-fold higher risk of death, fold higher risk of mortality when compared to those which echoed this study. In particular, our multivariate with D-dimer less than 0.5 μg/L. Potential mechanisms analysis showed that AST is an independent determinant include induction of procoagulant factors, systemic pro- of COVID-19 mortality. As a 13 kilo Dalton proteinase inflammatory cytokine responses which are mediators inhibitor from the cystatin super family of cysteine of atherosclerosis directly causing plaque rupture by protease inhibitors, Cystatin C plays a crucial impact in local inflammation, and haemodynamic changes, which intra-cellular catabolism of peptides and proteins (24). contribute to ischaemia and thrombosis. Acute myocardial Evidences have proven that this biomarker is a superior injury is a well-recognized cardiovascular complication indicator of kidney function as compared with creatinine among COVID-19 patients. Only 12% of survivors had (25,26). To date, influence of SARS-CoV-2 on cystatin C heart failure as opposed to 52% of the deceased (7). Several has not been reported. Our research revealed patients with recent researches pointed out that a greater concentration higher serum cystatin C were prone to develop a poorer of lactate dehydrogenase (LDH), N-terminal pro B-type prognosis. A study investigated the potential parameters natriuretic peptide (NT-proBNP), CK and cTnI was associated with imaging progression on chest CT from related to the mortality rate of COVID-19 (32-34). Wang COVID-19 cases and found serum cystatin C and creatinine et al. (12) reported the median CK-MB level was remarkably were significantly higher in imaging progression patients, lower in cases not requiring ICU care as compared with which means a more adverse prognosis (27). Chen et al. (28) those in ICU (14 vs. 18 U/L, P<0.01). Our result echoed reported that the comorbidity of renal disease at admission the previous studies that the elevated CK-MB was an and the progression of acute renal injury within the duration independent predictor of death in COVID-19 patients. of hospitalization in cases with COVID-19 is related to in- Direct myocardial injury owing to the effect of systemic hospital mortality. As elevatory hepatic and renal injury inflammation or viral myocarditis seems to be the most indicators are highly associated with the exacerbation common mechanisms of acute cardiac injury. Studies on and subsequent death risk, it is strongly recommended to MERS-CoV and SARS-CoV had indicated that fibrinolysis monitor the liver and renal function during hospitalization. and hyper coagulation could promote the likelihood of CRP is a non-specific acute-phase protein induced by formation of microthrombus and further aggravate the risk IL-6 in the liver and a sensitive biomarker of tissue damage, of organ failure (35). In this study, thrombocytopenia in the infection and inflammation. COVID-19 cases with elevated early stage was determined to a determinant of COVID-19 plasma CRP in the early stage are predisposed to develop mortality, which was consistent with other studies (36,37). severe illness and induce the death. CRP has been proposed As well known, hyper activity of fibrinolysis can lead to to be adopted as a prognostic indicator, and higher levels the accelerating consumption of platelet. On the other of CRP suggesting increased risk of disease deterioration hand, SARS-CoV-2 infection damaged lung tissue, causing (29,30). Our result supported the studies of CRP level with activation, aggregation and entrapment of the platelet. COVID-19 severity elsewhere. In this study, COVID-19 This induces thrombosis on the site of pulmonary injury, patients with CRP ≥100 mg/L had 2.830 (95% CI: which increases the depletion of platelet. Furthermore, 1.311–6.109) times greater risk of mortality, as compared virus infection might induce immunologic injury to platelet with patients with CRP <100 mg/L. Clinically, elevated by bringing about auto-antibodies and immune complexes. CRP level could be early marker of nosocomial infection Since reduced platelet count may aggravate the disease in COVID-19 patients that were difficult to recover, and and be related to the adverse disease progression and even empirical antibiotics treatment should be administered death, it is recommended to frequently measure the platelet early to prevent worsened prognosis. Nearly 90 percent count, leading to more effective measures much earlier. A of pneumonia inpatients presented elevated coagulation nomogram was constructed on the basis of the independent activity, characterized by higher D-dimer level (31). As predictors deriving from multivariate regression. The described previously, elevated D-dimer has been known nomogram incorporating variables of governing COVID-19 as a crucial determinant of mortality in COVID-19 cases mortality can serve as an easy means to evaluate the risk of

© Annals of Palliative Medicine. All rights reserved. Ann Palliat Med 2021;10(4):3937-3950 | http://dx.doi.org/10.21037/apm-20-2107 3948 Li et al. Determinants of mortality of patients with COVID-19 death for COVID-19 patients and may assist physicians in STROBE reporting checklist. Available at http://dx.doi. decision making concerning delivery of medical care and org/10.21037/apm-20-2107 optimal treatment. To our knowledge, this is the first report of the application of the nomogram in predicting the risk of Data Sharing Statement: Available at http://dx.doi. mortality of COVID-19 patients. org/10.21037/apm-20-2107 Our study had several limitations. Firstly, not all laboratory indicators were detected in all cases. Although missing values Conflicts of Interest: All authors have completed the ICMJE were replaced through multiple imputation technique, their uniform disclosure form (available at http://dx.doi. role in predicting mortality might be underestimated. Secondly, org/10.21037/apm-20-2107). The authors have no conflicts owing to massive missing of chest CT data at an early stage of of interest to declare. the epidemic, this study could not evaluate the forecasting role of chest CT abnormalities. Its role for predicting mortality risk Ethical Statement: The authors are accountable for all of COVID-19 has been proven by previous study (13). Thirdly, aspects of the work in ensuring that questions related regardless of the fact that the subjects came from 2 campuses to the accuracy or integrity of any part of the work are of a designated hospital, generalisability of the study results is appropriately investigated and resolved. The Ethics probably restricted by the sample size, and is necessary to be Committees of Wuhan Third Hospitals has approved validated on the basis of much larger patients. this study (No.: KY2020-007) and waived the individual informed consent due to the retrospective analysis. The study was carried out in conformity to the Declaration of Conclusions Helsinki (as revised in 2013). In this research, we identified that older age, comorbidity of cerebrovascular disease, higher WBC count, AST, cystatin Open Access Statement: This is an Open Access article C, CRP, CK-MB, D-dimer and lower platelet count on distributed in accordance with the Creative Commons admission as the independent determinants to predict the Attribution-NonCommercial-NoDerivs 4.0 International mortality risk in COVID-19 cases. Considering the ongoing License (CC BY-NC-ND 4.0), which permits the non- worldwide pandemic of COVID-19, the current research commercial replication and distribution of the article with will be helpful to identify patients at great risk of death at the strict proviso that no changes or edits are made and the an early stage and guide the clinical treatment. original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/. Acknowledgments

We acknowledge all health-care workers involved in the References diagnosis and treatment of patients in study site, and we would like to thank Dr. Binxun Liao, Dr. Yueyuan 1. Huang C, Wang Y, Li X, et al. Clinical features of patients Wang, Dr. Zhipeng Li for helping data cleaning. We also infected with 2019 novel coronavirus in Wuhan, China. acknowledge Prof. Genyan Yang (Centers for Disease Lancet 2020;395:497-506. Control and Prevention, Atlanta, GA, USA) for language 2. Chen N, Zhou M, Dong X, et al. Epidemiological editing and critical review of the manuscript. and clinical characteristics of 99 cases of 2019 novel Funding: This study was funded by grants from Shanghai coronavirus pneumonia in Wuhan, China:a descriptive Municipal Science & Technology Commission Key Scientific study. Lancet 2020;395:507-13. and Technological Innovation Action Plan (18411950900), 3. World Health Organization. Coronavirus disease National Nature Science Funds of China (81772107) and (COVID-19) weekly epidemiological update. 2020. Shanghai Shenkang Hospital Development Center Clinical Accessed 31 August 2020. Available online: https:// Science and Technology Innovation Project (SHDC12017116). www.who.int/docs/default-source/coronaviruse/ situation-reports/20200831-weekly-epi-update-3. pdf?sfvrsn=d7032a2a_4 Footnote 4. Wu Z, McGoogan JM. Characteristics of and Important Reporting Checklist: The authors have completed the Lessons From the Coronavirus Disease 2019 (COVID-19)

Outbreak in China: Summary of a Report of 72 314 17. Altschul DJ, Unda SR, Benton J, et al. A novel severity Cases From the Chinese Center for Disease Control and score to predict inpatient mortality in COVID-19 patients. Prevention. JAMA 2020;323:1239-42. Sci Rep 2020;10:16726. 5. Booth CM, Matukas LM, Tomlinson GA, et al. Clinical 18. Opal SM, Girard TD, Ely EW. The immunopathogenesis features and short-term outcomes of 144 patients with of sepsis in elderly patients. Clin Infect Dis SARS in the greater Toronto area. JAMA 2005;41:S504-12. 2003;289:2801-9. 19. Zheng Z, Peng F, Xu B, et al. Risk factors of critical & 6. Rahman A, Sarkar A. Risk factors for fatal Middle East mortal COVID-19 cases: A systematic literature review Respiratory Syndrome coronavirus infections in Saudi and meta-analysis. J Infect 2020;81:e16-25. Arabia: analysis of the WHO line list, 2013-2018. Am J 20. Li YC, Bai WZ, Hashikaw T. The neuroinvasive potential Public Health 2019;109:1288-93. of SARS-CoV2 may play a role in the respiratory failure of 7. Zhou F, Yu T, Du RH, et al. Clinical course and risk COVID-19 patients. J Med Virol 2020;92:552-5. factors for mortality of adult inpatients with COVID-19 21. Zhang Y, Zheng L, Liu L, et al. Liver impairment in in Wuhan, China: a retrospective cohort study. Lancet COVID-19 patients: a retrospective analysis of 115 cases 2020;395:1054-62. from a single centre in Wuhan city, China. Liver Int 8. Wu C, Chen X, Cai Y, et al. Risk factors associated with 2020;40:2095-103. acute respiratory distress syndrome and death in patients 22. Lei F, Liu YM, Zhou F, et al. Longitudinal association with corona virus disease 2019 pneumonia in Wuhan, between markers of liver injury and mortality in China. JAMA Intern Med 2020;180:934-43. Erratum in: COVID-19 in China. Hepatology 2020;72:389-98. JAMA Intern Med 2020;180:1031. 23. Parohan M, Yaghoubi S, Seraj A. Liver injury is associated 9. Livingston E, Bucher K. Coronavirus Disease 2019 with severe coronavirus disease 2019 (COVID-19) (COVID-19) in Italy. JAMA 2020;323:1335. infection: a systematic review and meta-analysis of 10. Guan WJ, Liang WH, Zhao Y, et al. Comorbidity and retrospective studies. Hepatol Res 2020;50:924-35. its impact on 1590 patients with COVID-19 in China: a 24. Nakhjavan-Shahraki B, Yousefifard M, Ataei N, et al. nationwide analysis. Eur Respir J 2020;55:2000547. Accuracy of cystatin C in prediction of acute kidney injury 11. Huang R, Zhu L, Xue LY, et al. Clinical findings of in children; serum or urine levels: which one works better? patients with coronavirus disease 2019 in Jiangsu Province, A systematic review and meta-analysis. BMC Nephrol China: a retrospective, multi-center study. PLoS Negl 2017;18:120. Trop Dis 2020;14:e0008280. 25. Dharnidharka VR, Kwon C, Stevens G. Serum systatin 12. Wang D, Hu B, Hu C, et al. Clinical Characteristics of C is superior to serum creatinine as a marker of kidney 138 Hospitalized Patients With 2019 Novel Coronavirus- function: a meta-analysis. Am J Kidney Dis 2002;40:221-6. Infected Pneumonia in Wuhan, China. JAMA 26. Roos JF, Doust J, Tett SE, et al. Diagnostic accuracy of 2020;323:1061-9. cystatin C compared to serum creatinine for the estimation 13. Liang W, Liang H, Ou L, et al. Development and of renal dysfunction in adults and children- a meta- Validation of a Clinical Risk Score to Predict the analysis. Clin Biochem 2007;40:383-91. Occurrence of Critical Illness in Hospitalized Patients 27. Yang Z, Shi J, He Z, et al. Predictors for imaging With COVID-19. JAMA Intern Med 2020;180:1081-9. progression on chest CT from coronavirus disease 2019 14. Feng Y, Ling Y, Bai T, et al. COVID-19 with Different (COVID-19) patients. Aging (Albany NY) Severity: A Multi-center Study of Clinical Features. Am J 2020;12:6037-48. Respir Crit Care Med 2020;201:1380-8. 28. Cheng Y, Luo R, Wang K, et al. Kidney disease is 15. Wynants L, Van Calster B, Collins GS, et al. Prediction associated with in-hospital Death of patients with models for diagnosis and prognosis of covid-19 COVID-19. Kidney Int 2020;97:829-38. infection: systematic review and critical appraisal. BMJ 29. Chen W, Zheng KI, Liu SD, et al. Plasma CRP level is 2020;369:m1328. positively associated with the severity of COVID-19. Ann 16. Fumagalli C, Rozzini R, Vannini M, et al. Clinical risk Clin Microbiol Antimicrob 2020;19:18. score to predict in-hospital mortality in COVID-19 30. Liu F, Li L, Xu MD, et al. Prognostic value of patients: a retrospective cohort study. BMJ Open interleukin-6, C-reactive protein, and procalcitonin in 2020;10:e040729. patients with COVID-19. J Clin Virol 2020;127:104370.

31. Milbrandt EB, Reader MC, Lee M, et al. Prevalence and 2020;46:846-8. significance of coagulation abnormalities in community- 35. So LK, Lau AC, Yam LY, et al. Development of a standard acquired pneumonia. Mol Med 2009;15:438-45. treatment protocol for severe acute respiratory syndrome. 32. Inciardi RM, Adamo M, Lupi L, et al. Characteristics Lancet 2003;361:1615-7. and outcomes of patients hospitalized for COVID-19 and 36. Lippi G, Plebani M, Henry BM. Thrombocytopenia cardiac disease in northern Italy. Eur Heart J is associated with severe cornonvirus disease 2020;41:1821-9. 2019(COVID-19) infections: A meta-analysis. Clin Chim 33. Gaze DC. Clinical Utility of Cardiac Troponin Acta 2020;506:145-8. Measurement in COVID-19 Infection. Ann Clin Biochem 2020;57:202-5. 37. Zhao X, Wang K, Zuo P, et al. Early decrease in blood 34. Ruan Q, Yang K, Wang W, et al. Clinical predictors of platelet count is associated with poor prognosis in mortality due to COVID-19 based on an analysis of data COVID-19 patients-indications for predictive, preventive, of 150 patients from Wuhan, China. Intensive Care Med and personalized medical approach. EPMA J 2020;11:1-7.

Cite this article as: Li J, Yang L, Zeng Q, Li Q, Yang Z, Han L, Huang X, Chen E. Determinants of mortality of patients with COVID-19 in Wuhan, China: a case-control study. Ann Palliat Med 2021;10(4):3937-3950. doi: 10.21037/apm-20-2107

Table S1 Reference range of laboratory indicators Indicator Unit Reference range

C-reactive protein mg/L <5.00

White blood cell count ×109/L 3.50–9.50

Neutrophil count ×109/L 1.80–6.30

Lymphocyte count ×109/L 1.10–3.20

Monocyte count ×109/L 0.12–0.80

Platelet count ×109/L 125–350

ALT U/L 9.0–50.0

AST U/L 15.0–40.0

Total bilirubin umol/L 3.40–20.50

Total protein g/L 60.0–80.0

Globulin g/L 20.0–30.0

Blood urea nitrogen mmol/L 3.60–9.50

Creatinine umol/L 57.00–111.00

Cystatin C mg/L 0.5–1.1

PT second 11–16

TT second 18–25

α-HBDH IU/L 90–182 Creatine Kinase U/L 30.00–200.00

CK-MB U/L <24

D-dimer μg/L <0.50 ALT, alanine aminotransferase; AST, aspartate aminotransferase; PT, prothrombin time; TT, thrombin time; α-HBDH, α-hydroxybutyrate dehydrogenase; CK-MB, creatine kinase isoenzyme.